School Work

Mindfulness and emotion regulation: The development and initial validation of the Cognitive and Affective Mindfulness Scale-Revised (CAMS-R)

Mindfulness and emotion regulation: The development and initial validation of the Cognitive and Affective Mindfulness Scale-Revised (CAMS-R)
of 14
All materials on our website are shared by users. If you have any questions about copyright issues, please report us to resolve them. We are always happy to assist you.
  J Psychopathol Behav Assess (2007) 29:177–190DOI 10.1007/s10862-006-9035-8 ORIGINAL PAPER Mindfulness and Emotion Regulation: The Developmentand Initial Validation of the Cognitive and AffectiveMindfulness Scale-Revised (CAMS-R) Greg Feldman · Adele Hayes · Sameet Kumar · Jeff Greeson · Jean-Philippe Laurenceau Published online: 7 November 2006 C  Springer Science + Business Media, LLC 2006 Abstract Asinterestgrowsinmindfulnesstrainingasapsy-chosocial intervention, it is increasingly important to quan-tify this construct to facilitate empirical investigation. Thegoal of the present studies was to develop a brief self-reportmeasure of mindfulness with items that cover the breadthof the construct and that are written in everyday language.The resulting 12-item measure demonstrated acceptable in-ternal consistency and evidence of convergent and discrimi-nant validity with concurrent measures of mindfulness, dis-tress, well-being, emotion-regulation, and problem-solvingapproaches in three samples of university students. To ad-dress potential construct contamination in two items, data Portions of this paper were previously presented in a symposium atthe annual convention of the Association for the Advancement of Behavior Therapy, Boston, MA in November of 2003.G. Feldman · J. GreesonDepartment of Psychology, University of Miami,P.O. Box 24-8073, Coral Gables, FL 33124, USAA. Hayes · J.-P. LaurenceauDepartment of Psychology, University of Delaware,108 Wolf Hall,Newark, Delaware 19716, USAS. Kumar Mt. Sinai Comprehensive Cancer Center,4300 Alton Rd 125, Miami, FL 33140, USAJ. GreesonPresent address: Department of Psychiatry and BehavioralSciences, Duke Center for Integrative Medicine,DUMC Box 3022, Durham, NC 27708, USAG. Feldman (  )Department of Psychology, Simmons College,300 the Fenway, Boston, MA 02114, USAe-mail: are also presented on an alternate 10-item version of themeasure. Keywords Mindfulness.Emotion regulation.Depression.AnxietyThe field of clinical psychology has witnessed a recent surgeofinterestinmindfulnesstrainingasapsychosocialinterven-tion. Two recent meta-analyses document the evidence for mindfulness training as an efficient method of distress re-duction in medical and psychiatric populations, as well as innon-clinical populations (Baer,2003;Grossman, Niemann, Schmidt, & Walach,2004). The development of a reliableand valid self-report measure of mindfulness is an importantresearch priority to further advance the scientific study of this construct (Bishop et al.,2004; Dimidjian & Linehan,2003).A critical task in measurement development is to gener-ate an operational definition of the mindfulness construct.In Kabat-Zinn’s writings on mindfulness (e.g., 2003), hedefines the term as “the awareness that emerges throughpaying attention on purpose, in the present moment, andnon-judgmentally to the unfolding of experience momentto moment.” In an effort to facilitate measurement develop-ment and hypothesis-testing of the construct, a consensuspanel was recently convened to provide a common opera-tional definition for researchers. The panel (Bishop et al.,2004) generated a definition that emphasizes the regulationof attention (“the self-regulation of attention so that it ismaintained on immediate experience, thereby allowing for increased recognition of mental events in the present mo-ment”) and one’s orientation to experience (“adopting a par-ticular orientation towards one’s experiences in the presentmoment, an orientation that is characterized by curiosity,openness, and acceptance”). Common to these definitions Springer   178 J Psychopathol Behav Assess (2007) 29:177–190 are four components: 1) the ability to regulate attention, 2)an orientation to present or immediate experience, 3) aware-ness of experience, and 4) an attitude of acceptance or non- judgment towards experience.Several years ago, the first three authors of the presentseries of studies began work on a self-report measure of individual differences in mindfulness called the Cognitiveand Affective Mindfulness Scale (CAMS; Kumar,2005;Kumar, Feldman, & Hayes,2005). When the CAMS wascreated, no other self-report measures of mindfulness wereavailable. Since that time, there are four additional self-report measures of mindfulness in various stages of de-velopment. It is useful to have multiple measures in theearly stages of operationalization, as the existing measurescome from slightly different perspectives, emphasize dif-ferent components of mindfulness, and sample a varietyof populations. Having some choice in measures allowsresearchers flexibility in matching instruments to researchobjectives.The CAMS (Kumar et al.,2005)consists of 18 items designed to capture a broad conceptualization of mindful-ness (Kabat-Zinn,1990)with language that is not specific to any particular type of meditation training, which allowsfor use with a range of samples. The measure was firsttested in a small sample of depressed individuals receiv-ing an integrative psychotherapy that includes mindfulnesstraining (Hayes, Beevers, Feldman, Laurenceau, & Perlman,2005). The CAMS was found to be sensitive to change(Kumar et al.,2005) and demonstrated concurrent validity(Feldman,Kumar,Galyardt,&Hayes,2002),buttheinternal consistency was low and the items generally lacked directassessment (i.e., forward-scored items) of key aspects of theconstruct. The factor structure of the CAMS was not testedin this small sample. Subsequent exploratory factor analy-sesinlargestudentsamplesgenerallysupportedafour-factor structureconsistentwiththeory,butprimaryloadingsofspe-cific items failed to replicate across samples (Feldman et al.,2005). The present study was designed to address these lim-itations and to test this revised measure against a broader range of criterion measures.Like the CAMS, the Mindful Attention Awareness Scale(MAAS; Brown & Ryan,2003) assesses mindfulness withitems designed to be free of specialized, metaphorical, andidiomatic language. The MAAS includes the attentional andawareness aspects of mindfulness, but not the attitudinalcomponentsofacceptanceandnon-judgmentthatareempha-sized across mindfulness-based clinical interventions (Baer,2003),potentiallylimitingtheuseoftheMAASinclinically-related research. Like the CAMS, the Freiburg MindfulnessInventory (FMI; Buchheld, Grossman, & Walach,2002) andthe Kentucky Inventory of Mindfulness Skills (KIMS; Baer,Smith, & Allen,2004) assess a broad conceptualizationof mindfulness. The FMI is limited in that it is intendedfor use only with individuals who already have familiar-ity with the principles of mindfulness. The KIMS can bescored to produce four internally-consistent subscales; how-ever, a disadvantage is its length, which is more than twiceas long as either the MAAS or the CAMS. Shorter mea-sures can be useful when respondent burden is a concern.The Toronto Mindfulness Scale (TMS;Lau et al., in press)is a 13-item measure of the state of mindful self-regulationof attention and approach to experience. This measure isdesigned to be administered immediately after a medita-tion session, which limits the settings in which it can beapplied.The four existing measures reviewed above (i.e., FMI,MAAS,KIMS,andTMS)andtheearlyversionoftheCAMSoffer unique advantages and disadvantages in terms of their conceptualcoverageofthecomponentsofmindfulness,their item content, their length, and their generalizability. TheCAMS offers the advantages of capturing a multi-facetedconceptualization of mindfulness, being relatively brief, andusing language and a format that does not restrict its use toa specific setting (e.g., mindfulness meditation training). InundertakingarevisionoftheCAMS,thegoalsweretwofold.First,furtheranalysesofthemeasure’spsychometricproper-ties were conducted and, where appropriate, its item contentwasrefined.Second,consistentwiththestrengthsoftheorig-inal CAMS, a measure was created that has comprehensiveconceptual coverage, uses clear everyday language, and isbrief.Theoretical writings on mindfulness, both modern andancient, describe the practice of mindfulness as a meansto quiet the mind, decrease suffering, and enhance qualityof life (Gampopa,2000; Kabat-Zinn,2003; Walshe,1987). Individuals who report lower levels of mindfulness alsotend to report more distress and less psychological well-being (Baer et al.,2004;Brown & Ryan,2003;Feldman et al.,2002). In the present study, it was hypothesized thatthe revised CAMS (CAMS-R) would demonstrate a sim-ilar pattern of associations with measures of distress andwell-being.In traditional Buddhist writings, mindfulness is thoughtto improve well-being by reducing tendencies towards aver-sion and attachment to internal and external phenomena,thereby facilitating emotional regulation (Kumar,2002). Non-mindful approaches to internal experience have beencharacterized as under- or over-engagement with internalexperiences (Buchheld et al.,2002; Hayes & Feldman,2004; Kabat-Zinn,1990). Emotional under-engagement has been studied in literatures on experiential avoidance(Hayes, Wilson, Gifford, Follette, & Strosahl,1996) andthought suppression (Wegner,1994). Over-engagement in- cludesprocessesinwhichindividualsexaggerateorelaborateupon initial symptoms of distress, such as worry (Borkovec,1994), rumination (Nolen-Hoeksema & Morrow,1991), and Springer   J Psychopathol Behav Assess (2007) 29:177–190 179 overgeneralization (Carver,1998). 1 In contrast to under- andover-engagement, mindfulness involves a process of bothacknowledging emotions as transient phenomena and fullyexperiencingemotionswithoutnecessarilyactinguponthem(Kabat-Zinn,1990). Higher levels of self-reported mind- fulness have been associated with less over- and under-engagement, such as experiential avoidance, thoughtsuppression, and rumination, and with higher levels of emo-tional intelligence (Baer et al.,2004; Brown & Ryan,2003; Feldman et al.,2002). Thus, it was hypothesized that higher  mindfulness scores on the CAMS-R would be associatedwith less under-engagement (experiential avoidance andthought suppression), less over-engagement (worry, ru-mination, and overgeneralization), and more emotionalintelligence.Theoretical accounts highlight cognitive, attentional, andbehavioral flexibility as components and consequencesof mindfulness (Bishop et al.,2004;Borkovec,2002; Kabat-Zinn,1990;Roemer & Orsillo,2003). The opposite pattern of taking a narrow perspective and responding in anautomatic, habitual, and inflexible manner is often associ-ated with poor problem-solving and more distress. 2 Thus,it was hypothesized that self-reported flexibility in copingwould be associated with higher mindfulness scores on theCAMS-R. In the practice of mindfulness, individuals are of-ten reminded to refrain from generating ‘to do’ lists as theymeditate. Therefore, one aspect of the coping process thatis particularly relevant to mindfulness is that of mental an-ticipatory coping (Aspinwall & Taylor,1997; Feldman &Hayes,2005), or the process by which individuals mentallyprepareforpendingproblems.Itwashypothesizedthatmind-fulness would be associated with adaptive forms of mentalpreparation, such as problem analysis and mental rehearsalof action, but that these associations would not be strong, asmindfulnessisneitherananalyticor‘doing’modeofthought 1 There is considerable evidence that under- and over-engagement canand do co-occur. Individuals with little skill in regulating emotionsmay vacillate between under- and over-engagement. In addition, indi-viduals can use over-engagement as a form of avoidance. Excessiveworry may be a means of distracting oneself from more upsetting top-ics (Borkovec,1994). Similarly, rumination has been conceptualizedas a means of distracting one’s self from one’s actual life problems(Jacobson, Martell, & Dimidjian,2001). There is also recent evidence that thought suppression prospectively predicts increased ruminationfor individuals experiencing a high degree of life stress (Wenzlaff &Luxton,2003). Thus, under- and over-engagement are not necessarilymutually exclusive. However, these broad categorizations are used hereto organize the criterion measures and the discussion of them. 2 In this way, there is considerable overlap with an alternative concep-tualization of mindfulness developed by Ellen Langer and colleagues(Langer,1989), which involves both paying attention to one’s envi-ronment, actively viewing situations from multiple perspectives, andresponding in novel ways. This conceptualization, however, departssomewhat from Eastern notions of mindfulness that inform clinicalinterventions, which tend to emphasize “allowing” thoughts to passthrough the mind rather than actively manipulating them. (Teasdale,1999). Individuals low in mindfulness would tendto unproductively dwell on future problems (a form of over-engagement) or passively fantasize about their resolution,while ignoring stressful details of the coping process (a formof under-engagement).Inthefollowingsections,theprocessbywhichitemsweredeveloped and selected for the revised Cognitive and Affec-tiveMindfulnessScale(CAMS-R)ispresented(Study1)andevidence of convergent and discriminant validity is reportedusing a student sample (Study 2). Study 1 Thegoalofthefirststudywastodesignameasureofmindful-nessthatisbriefandyetcaptures thebreadth oftheconstructofmindfulness.Assuch,thegoalwasnottocreateameasureofsufficientlengthtoyieldindependentsubscaleswithsuffi-cientlystrongseparatepsychometricpropertiestoassesssep-arate aspects of the construct. Instead, this study tests the hy-pothesis that a single, higher-order construct of mindfulness(i.e., second-order factor) can be inferred from four compo-nents (i.e., four first-order factors: attention, present-focus,awareness, and acceptance). A common pitfall in measuredevelopment is the “attenuation paradox” in which internalconsistency of a measure is artificially increased by includ-ing items that access a single, narrow aspect of a broader construct (John & Benet-Martinez,2000). The purpose of these analyses is to ensure that the mindfulness total scorein the CAMS-R adequately represents the four componentsof mindfulness identified in previous operational definitions(Bishop et al.,2004;Kabat-Zinn,2003). Structural equation modeling (SEM) was used to guidethe early phase of item-selection. For instrument develop-ment, use of SEM factor analytic strategies has two ma- jor advantages over traditional exploratory factor analysis(Bollen,1989; Kline,2004). First, unlike exploratory factor  analysis, which is largely data-driven, SEM models allowthe researcher to test a priori hypotheses about the under-lying structure of the items. Second, SEM allows for theresearcher to simultaneously test for the presence of bothfirst- and second-order factors, as proposed in this model.Method  Item generation A pool of 35 items was generated by a group of researcherswith expertise in mindfulness meditation, emotion regula-tion, and questionnaire development. Items were informedbyareviewofphilosophicalwritingsonmindfulness,aswellas writings on clinical applications of mindfulness. Itemswere written to convey attitudes and approaches towardsinternal experiences of emotions and thoughts. Items that Springer   180 J Psychopathol Behav Assess (2007) 29:177–190 reflectresponsestoexternalexperiencesorbodilysensationswere not included, as these concerns are not relevant acrossdisorders or treatment modalities. This item pool containedthe 18 items from the srcinal CAMS (Kumar et al.,2005).However,asnotedpreviously,preliminaryexploratoryfactor analyses of the CAMS revealed that the primary loadings of several items were unstable across samples (Feldman et al.,2005).Thus,forthepresentstudy,17newitemswerewrittenand added to the item pool with the goal of more preciselycapturing the four factors suggested by theory and prelim-inary exploratory factor analyses. The item pool containedboth forward- and reverse-scored items. All items in thepresent study were designed to be comprehensible to indi-viduals with no prior experience with mindfulness practiceor meditation.  Participants An ethnically-diverse sample of 548 university studentsresponded to this item set in large testing sessionsduring two semesters in partial fulfillment of the re-search participation component of their introductory psy-chology course. Data from 250 students (Sample 1,64.2% women, 35.8% men; mean age = 19.31, SD = 2.66;Ethnicity: 55.7% White/Caucasian, 18.4, Hispanic/ Latino, 8.8% Black/African American, 6.1% Asian-American, 11.0% Other/Mixed Heritage) were used totest the preliminary models, and 298 (Sample 2, 60.5%women, 39.5% men; mean age = 18.74, SD = 1.92; Eth-nicity: 55.7% White/Caucasian, 20.8, Hispanic/Latino,7.2%Black/AfricanAmerican,6.8%Asian-American,9.5%Other/MixedHeritage)werereservedforaconfirmatoryfac-tor analysis.Participants were asked to respond to the item set andwere given the following prompt “People have a variety of ways of relating to their thoughts and feelings. For each of the items below, rate how much each of these ways appliesto you .” Participants were asked to rate their responses on aLikertscalewiththefollowingoptions:1(  Rarely/Notatall ),2 ( Sometimes ), 3 ( Often ), or 4 (  Almost always ). Testing ses-sionswerepartofdatacollectionproceduresapprovedbytheUniversity of Miami Institutional Review Board and partic-ipants signed informed consent forms before participating.The CAMS-R was administered along with additional ques-tionnaires not relevant to the present study. The order of thequestionnaires was varied across testing sessions.Results  Item reduction and preliminary model testing A correlation matrix of the srcinal pool of 35 items was ex-amined to guide the first stage of item reduction. Items thattended to have low correlations with non-redundant itemswere eliminated. Twenty items were retained for model de-velopmentusingconfirmatoryfactoranalysis(CFA).Allfac-tor models were tested with the structural equation modelingsoftware program LISREL 8.51 (J¨oreskog & S¨orbom,2001)using maximum likelihood estimation. For all tested mod-els, the following indices were used to assess goodness of fit: Comparative Fit Index (CFI), root mean squared error of approximation (RMSEA), standardized root-mean squaredresidual (SRMR), and the chi-square statistic ( χ 2 ). Cut-off values (CFI “close to” .95, RMSEA “close to” .06, SRMR“close to” .08) were selected based on the recommendationsby Hu and Bentler (1999, p. 1).The twenty items that were retained from the initial itemreduction were divided into four theoretically-derived cat-egories (four attention items, four present-focus items, fiveawarenessitems,sevenacceptanceitems).ACFAmodelwastested with one second-order latent factor (mindfulness) andfour first-order latent factors, each indicated by the itemsthat reflect that construct. All four factors were tested si-multaneously. The 20-item model demonstrated relativelypoor fit [ χ 2 (160) = 388.09, p < .00001; RMSEA = .073;SRMR = .075; CFI = .85].After exploring the patterns of item loadings in the 20-item model, items were eliminated in a series of itera-tive factor models. Items were eliminated if they demon-strated poor loadings on their hypothesized factor, demon-strated high cross-loadings on non-hypothesized factors, or were judged to be redundant with other retained items. Alltested models were congeneric, which means that all itemswere allowed to load only on a single first-order factor andmeasurement error of the items was not permitted to becorrelated.Through this iterative process, a final model emergedthat was a good fit to the data [ χ 2 (50) = 81.04, p = .004;RMSEA = .050; SRMR = .051; CFI = .95]. This model ap-pears in Fig.1. The RMSEA, SRMR, and the CFI werewithin the cut-off range recommended by Hu and Bentler (1999). The chi-square test fell short of non-significance (anindication of model fit); however, it should be noted that thechi-square statistic has been criticized as an overly-sensitivetest that can suggest rejecting potentially useful models, par-ticularly as N becomes large (Bollen,1989). Consistent with the initially proposed model, this modelconsisted of one second-order latent factor (mindfulness)and four first-order latent factors (attention, present-focus,awareness, and acceptance). The 12 retained items appear inTable1along with their descriptive statisticsacross samples.Six of the retained items were from the srcinal CAMS. Inthe final model, each first-order factor had three items, thussatisfying the recommended minimum number of observedvariables per latent variable and latent variables per second-order factor (Kline,2004; Wegener & Fabrigar,2000). All Springer   J Psychopathol Behav Assess (2007) 29:177–190 181 0.390.340.460.870.89 0.530.350.900.82 Mindfulness AttentionPresentFocusAwareness Acceptance 1 6 122 7 115 8 93 4 10 0.550.890.840.730.810.740.780.32 0.810.420.33 0.360.680.330.570.71 0.490.890.670.190.470.690.31 Fig. 1 Standardized parameter estimates for a second-order confirmatory factor analysis(LISREL) for the Cognitive andAffective Mindfulness Scale – Revised (CAMS-R). Numbersin boxes correspond toCAMS-R questionnaire itemspresented in Table1 items had standardized loadings on their first-order factorsabove.30(range = .32to.81).Theitemswithloadingsinthe.30 to .40 range were retained because they added theoret-ically substantive breadth to the mindfulness factor model.Moreover, smaller factor loadings for items can occur whenattempting to create a broadband construct such as mindful-ness (Hoyle,2000). Cross-validation Because the model pruning approach can result in a modelthat achieves acceptable fit by capitalizing on chance error,an additional confirmatory factor analysis was performed ondatadrawnfromanindependentsample(Sample2,N = 298)to test the replicability of the model derived from the prelim-inary analyses. Maximum likelihood estimation in LISREL8.51 (J¨oreskog & S¨orbom,2001) was again used. The fitindices suggest that the correspondence between the pro-posed model and the data was acceptable [ χ 2 (50) = 110.58,  p < .0001, RMSEA = .064, SRMR = .052, CFI = .92]. Ingeneral, the loading pattern for the second sample was simi-lartothatofthefirstsample.Becauseviolationsofnormalitycanartificiallyinflatefitstatistics(Hu&Bentler,1999),skewand kurtosis were examined for each of the 12 retained itemsin all samples (See Table1). In all cases, these statisticsindicate that the data did not significantly depart from the as-sumption of normally distributed population scores (Kline,2004)asallskewvalues < |1.0|andallkurtosisvalues < |1.1|.  Internal consistency and factor intercorrelation After appropriate scoring reversals, the 12 retained itemsdemonstrated an acceptable level of internal consistencyin both samples. (Sample 1 α = .74; Sample 2 α = .77).Table2presents factor intercorrelations and internal con-sistency. Because the alpha coefficient is influenced by thenumber of items included in the measure (John & Benet-Martinez,2000), it is not surprising that three of the four  three-item subfactors had low internal consistency. This isparticularly true when the items of a brief subscale sam-ple the breadth of a construct (cf. Ryff & Keyes,1995). The magnitude of the intercorrelations between the latentvariables, calculated in a CFA in which the four latent vari-ables were allowed to covary, ranged from medium to large(Cohen,1988). The correlations between the unweighted scale scores ranged from small to medium.DiscussionOn the basis of results of preliminary and confirmatory SEMmodels, 12 items were retained for a measure of mindful ap-proachestothoughtsandfeelings,whichwascalledtheCog-nitive and Affective Mindfulness Scale–Revised (CAMS-R). This 12-item measure was shown to adequately samplethe four domains of mindfulness (attention, present-focus,awareness, acceptance/non-judgment). The overall CAMS-R, but not the subscales, demonstrated acceptable levels of internal consistency. These results are consistent with thefour components emphasized in several definitions of mind-fulness (Bishop et al.,2004;Kabat-Zinn,1990,2003). The results of this confirmatory factor analysis of the CAMS-Rwere recently replicated by another team of researchers in acommunity sample of adults diagnosed with inflammatorybowel disease (McPhail et al.,2005).The findings from Study 1 offer support for use of a sin-gle total mindfulness score rather than four subscale scores.First-order factor covariances were medium to large, whichsuggests considerable interrelationship between the con-structs assessed by the subscales. An additional advantageof using the total score rather than the first-order factors is Springer 
Related Search
We Need Your Support
Thank you for visiting our website and your interest in our free products and services. We are nonprofit website to share and download documents. To the running of this website, we need your help to support us.

Thanks to everyone for your continued support.

No, Thanks

We need your sign to support Project to invent "SMART AND CONTROLLABLE REFLECTIVE BALLOONS" to cover the Sun and Save Our Earth.

More details...

Sign Now!

We are very appreciated for your Prompt Action!